| Program ID | Capacity | Min Age | Max Age | ZIP Code | Latitude | Longitude | |
|---|---|---|---|---|---|---|---|
| count | 227746.000000 | 2.207110e+05 | 227746.000000 | 227746.000000 | 220353.000000 | 219615.000000 | 219615.000000 |
| mean | 150406.844046 | 5.655422e+03 | 8.686647 | 44.162668 | 60629.190104 | 41.851651 | -87.680025 |
| std | 36469.011260 | 7.016113e+05 | 6.518894 | 42.095325 | 27.371589 | 0.099947 | 0.118872 |
| min | 76358.000000 | 0.000000e+00 | 0.000000 | 0.000000 | 60018.000000 | 38.922466 | -120.961998 |
| 25% | 118969.250000 | 1.000000e+01 | 3.000000 | 12.000000 | 60617.000000 | 41.776459 | -87.717003 |
| 50% | 148861.500000 | 1.500000e+01 | 6.000000 | 18.000000 | 60628.000000 | 41.863098 | -87.680382 |
| 75% | 184012.750000 | 2.800000e+01 | 13.000000 | 99.000000 | 60641.000000 | 41.945400 | -87.638603 |
| max | 211184.000000 | 9.910181e+07 | 65.000000 | 171.000000 | 66210.000000 | 42.147499 | -87.530502 |
Project Report
Screenagers
Abstract
Following the start of the COVID-19 Pandemic, My CHI. My Future. launched with a goal of connecting Chicago youth to different opportunities. In this project, we sought to find gaps in equitability across program types based on different factors. We investigated accessibility to programs in the directory based on student income levels and minority statuses, accessibility to bilingual/multilingual programs, and accessibilities to programs in high crime areas. We suggest working with schools with higher percentages of Hispanic students to provide Academic Support Programs close in proximity to the students as well as ensuring these programs begin after school. Additionally, we advise the application to include a “Language” option where organizers can specify the languages of the opportunities while further advice into how to increase bilingual/multilingual opportunities based on the poor linguistic representation of the Chicago population found from the offered opportunities is also discussed. Lastly, we encourage further investment in youth-oriented programs in high-crime areas, as they can be beneficial and have a potential protective effect. With this, there should be a focus on the quality, type, and location of programs rather than quantity, as increasing the number of programs with strategic planning has the potential to reduce crime.
Note: This project deadline was extended by one day in return for three late days among the contributors.
1 Problem statement
We sought to analyze the equitability of access to programs in the My Chi. My Future. directory based on three different conditions: income levels and minority status, bilingualism/multilingualism, and crime rates.
We wanted to find out if Academic Support Programs were accessible to students attending schools with higher percentages of low income students as well as higher percentages of minority students.
Considering that about 35% of the Chicago population speaks a language other than English at home, we sought to find the language distribution of the opportunities offered in Chicago. In specific we looked to analyze the equitability of the bilingual/multilingual opportunities listed in the directory based on quantity, geographic location and language distribution.
Lastly, we wanted to analyze the relationship between the crime rate of an area and the number of programs available there. We wanted to see if program availability (and, more specifically, youth programs) has a role in reducing crime in historically high-crime areas, and what programs are most popular in low crime areas compared to high crime areas.
2 Data sources
This data set consists of many different youth programs Chicago between the years 2020 and 2024.
This data set contains information on Chicago Public Schools for the 2020-21 school year.
Language Spoken at Home and Ability to Speak English
Taken from the 2015-2019 American Community Survey, this table contains the population that speaks a certain language other than English and those that speaks only English at home per neighborhood in Chicago.
Records the crimes from 2001 to 2022 and contains the geographic location of where the crime occurred in Chicago.
3 Data quality check / cleaning / preparation
3.1 Distribution of Variables
My CHI. My Future. Data
| Missing Values | Unique Values | Value Counts | |
|---|---|---|---|
| Program Name | 0 | 30491 | {'Ice Skating - Freestyle Ice (Studio Rink) at... |
| Description | 0 | 42775 | {'Designated practice time for figure skaters ... |
| Org Name | 0 | 462 | {'Chicago Park District': 133258, 'Chicago Pub... |
| Category Name | 1 | 23 | {'Sports + Wellness.': 102187, 'Music & Art.':... |
| Address | 7316 | 1601 | {'3843 N. California Ave.': 11691, '810 E. 103... |
| City | 6196 | 27 | {'Chicago': 221475, nan: 6196, 'River Forest':... |
| State | 6196 | 3 | {'IL': 221368, nan: 6196, 'Illinois': 180, 'KS... |
| Program Type | 0 | 1 | {'workshop': 227746} |
| Program URL | 4358 | 110296 | {nan: 4358, 'https://youthreadychicago.cityspa... |
| Online Address | 218995 | 7030 | {nan: 218995, 'http://www.chicagoparkdistrict.... |
| Registration URL | 10005 | 107563 | {nan: 10005, 'https://youthreadychicago.citysp... |
| Registration Open | 225560 | 344 | {nan: 225560, '07/01/2023': 46, '05/15/2021': ... |
| Registration Deadline | 165833 | 901 | {nan: 165833, '03/24/2023': 1849, '03/23/2023'... |
| Start Date | 0 | 1333 | {'04/08/2024': 2283, '04/09/2024': 2274, '06/2... |
| End Date | 0 | 1302 | {'03/22/2024': 2051, '08/04/2023': 1994, '03/2... |
| Start Time | 24427 | 161 | {nan: 24427, '16:00': 17308, '10:00': 14371, '... |
| End Time | 24444 | 276 | {nan: 24444, '17:00': 14868, '18:00': 13098, '... |
| Contact Name | 14418 | 2425 | {'Managed Facilities': 29940, 'Community Recre... |
| Contact Email | 26328 | 1347 | {'play@chicagoparkdistrict.com': 133083, nan: ... |
| Contact Phone | 26726 | 1090 | {nan: 26726, '(773) 478-2609': 11691, '(312) 7... |
| Program Price | 0 | 4 | {'Free': 125468, '$50 or Less': 72715, 'More T... |
| Geographic Cluster Name | 11136 | 89 | {'Northwest Equity Zone': 21590, 'West Equity ... |
| Participants Paid | 2914 | 2 | {'Not Paid': 223956, nan: 2914, 'Paid, Type Un... |
| Transport Provided | 3253 | 2 | {False: 224336, nan: 3253, True: 157} |
| Has Free Food | 1750 | 2 | {False: 223543, True: 2453, nan: 1750} |
| Meeting Type | 0 | 2 | {'face_to_face': 217960, 'online': 9786} |
| Image | 215121 | 5130 | {nan: 215121, 'https://cityoflearning-uploads.... |
| Custom Categories | 221126 | 15 | {nan: 221126, 'Summertime CHI': 3632, 'Spring ... |
| Tag | 0 | 4 | {'Event': 120638, 'Program': 106251, 'Resource... |
| Location | 8131 | 1791 | {'POINT (-87.697303772 41.951400757)': 9307, n... |
Chicago Public Schools Data
| School_ID | Legacy_Unit_ID | Finance_ID | Zip | Student_Count_Total | Student_Count_Low_Income | Student_Count_Special_Ed | Student_Count_English_Learners | Student_Count_Black | Student_Count_Hispanic | ... | Student_Count_Hawaiian_Pacific_Islander | Student_Count_Ethnicity_Not_Available | Average_ACT_School | Mean_ACT | College_Enrollment_Rate_School | College_Enrollment_Rate_Mean | Graduation_Rate_School | Graduation_Rate_Mean | School_Latitude | School_Longitude | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| count | 655.000000 | 655.000000 | 655.000000 | 655.000000 | 655.000000 | 655.000000 | 655.000000 | 655.000000 | 655.000000 | 655.000000 | ... | 655.000000 | 655.000000 | 0.0 | 0.0 | 165.000000 | 165.0 | 141.000000 | 141.0 | 655.000000 | 655.000000 |
| mean | 569354.789313 | 5106.381679 | 36150.267176 | 60630.348092 | 503.621374 | 349.986260 | 74.438168 | 103.682443 | 181.512977 | 234.577099 | ... | 0.735878 | 2.041221 | NaN | NaN | 57.155152 | 67.2 | 73.342553 | 78.9 | 41.841401 | -87.677778 |
| std | 83074.064760 | 2534.237481 | 17548.073865 | 22.687433 | 402.438409 | 279.149048 | 55.878763 | 135.761314 | 198.149781 | 310.724929 | ... | 2.407718 | 10.506870 | NaN | NaN | 25.298639 | 0.0 | 24.119199 | 0.0 | 0.089028 | 0.058023 |
| min | 400009.000000 | 1010.000000 | 0.000000 | 60602.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | ... | 0.000000 | 0.000000 | NaN | NaN | 0.000000 | 67.2 | 0.000000 | 78.9 | 41.653663 | -87.841041 |
| 25% | 609734.500000 | 2955.000000 | 23796.000000 | 60618.000000 | 267.000000 | 178.000000 | 38.000000 | 3.000000 | 26.500000 | 11.000000 | ... | 0.000000 | 0.000000 | NaN | NaN | 39.700000 | 67.2 | 68.900000 | 78.9 | 41.771150 | -87.716862 |
| 50% | 609964.000000 | 4913.000000 | 26301.000000 | 60626.000000 | 408.000000 | 276.000000 | 62.000000 | 35.000000 | 137.000000 | 93.000000 | ... | 0.000000 | 0.000000 | NaN | NaN | 63.200000 | 67.2 | 80.400000 | 78.9 | 41.845967 | -87.678195 |
| 75% | 610181.000000 | 7080.000000 | 47056.000000 | 60640.000000 | 625.500000 | 439.000000 | 91.500000 | 177.000000 | 274.500000 | 385.000000 | ... | 1.000000 | 1.000000 | NaN | NaN | 77.500000 | 67.2 | 88.700000 | 78.9 | 41.910280 | -87.639971 |
| max | 610597.000000 | 9935.000000 | 70241.000000 | 60827.000000 | 4382.000000 | 2669.000000 | 535.000000 | 747.000000 | 1961.000000 | 2573.000000 | ... | 31.000000 | 203.000000 | NaN | NaN | 93.700000 | 67.2 | 99.100000 | 78.9 | 42.021091 | -87.527985 |
8 rows × 26 columns
| Missing Values | Unique Values | Value Counts | |
|---|---|---|---|
| Short_Name | 0 | 655 | {'PROVIDENCE ENGLEWOOD': 1, 'PATHWAYS - AVONDA... |
| Long_Name | 0 | 655 | {'Providence Englewood Charter School': 1, 'Pa... |
| Primary_Category | 0 | 3 | {'ES': 471, 'HS': 176, 'MS': 8} |
| Summary | 9 | 637 | {nan: 9, 'LEARN creates an intimate, resource-... |
| Administrator_Title | 0 | 2 | {'Principal': 536, 'Director': 119} |
| ... | ... | ... | ... |
| Network | 0 | 22 | {'Charter': 97, 'ISP': 82, 'Options': 36, 'Net... |
| Is_GoCPS_Elementary | 1 | 2 | {True: 427, False: 227, nan: 1} |
| Open_For_Enrollment_Date | 0 | 19 | {'09/01/2004': 516, '07/01/2012': 32, '07/01/2... |
| Closed_For_Enrollment_Date | 649 | 2 | {nan: 649, '06/30/2020': 4, '06/30/2021': 2} |
| Location | 0 | 654 | {'POINT (-87.6601387 41.8627144)': 2, 'POINT (... |
61 rows × 3 columns
Language Spoken at Home and Ability to Speak English
| TOT_POP | LING_ISO | ENGLISH | SPANISH | SLAVIC | CHINESE | TAGALOG | ARABIC | KOREAN | OTHER_EURO | |
|---|---|---|---|---|---|---|---|---|---|---|
| count | 77.000000 | 77.000000 | 77.000000 | 77.000000 | 77.000000 | 77.000000 | 77.000000 | 77.000000 | 77.000000 | 77.000000 |
| mean | 35175.324675 | 4829.142857 | 21082.389610 | 7951.675325 | 962.207792 | 623.311688 | 268.051948 | 235.272727 | 105.493506 | 1045.662338 |
| std | 23094.415237 | 5640.808466 | 16785.046601 | 10730.523932 | 1615.527365 | 1640.672602 | 482.862024 | 500.654063 | 173.003774 | 1779.184294 |
| min | 2006.000000 | 45.000000 | 1884.000000 | 39.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 |
| 25% | 18933.000000 | 337.000000 | 8809.000000 | 627.000000 | 37.000000 | 8.000000 | 0.000000 | 8.000000 | 0.000000 | 106.000000 |
| 50% | 29936.000000 | 3214.000000 | 17769.000000 | 3385.000000 | 259.000000 | 71.000000 | 53.000000 | 52.000000 | 17.000000 | 309.000000 |
| 75% | 45909.000000 | 8156.000000 | 26523.000000 | 12276.000000 | 1307.000000 | 422.000000 | 259.000000 | 226.000000 | 144.000000 | 1364.000000 |
| max | 101316.000000 | 26267.000000 | 80159.000000 | 54234.000000 | 9371.000000 | 10778.000000 | 2457.000000 | 3820.000000 | 687.000000 | 12650.000000 |
| Unique Values | Missing | Value Counts | |
|---|---|---|---|
| community | 77 | 0 | {'DOUGLAS': 1, 'SOUTH DEERING': 1, 'BRIGHTON P... |
Chicago Crime Data
| ID | Beat | District | Ward | Community Area | X Coordinate | Y Coordinate | Year | Latitude | Longitude | |
|---|---|---|---|---|---|---|---|---|---|---|
| count | 2.396200e+05 | 239620.000000 | 239620.000000 | 239610.000000 | 239620.000000 | 2.348850e+05 | 2.348850e+05 | 239620.0 | 234885.000000 | 234885.000000 |
| mean | 1.273171e+07 | 1154.060375 | 11.310892 | 23.384517 | 36.269427 | 1.165381e+06 | 1.887039e+06 | 2022.0 | 41.845613 | -87.668600 |
| std | 7.082832e+05 | 707.912519 | 7.075574 | 14.210173 | 21.554607 | 1.679381e+04 | 3.229561e+04 | 0.0 | 0.088833 | 0.061010 |
| min | 2.654300e+04 | 111.000000 | 1.000000 | 1.000000 | 1.000000 | 0.000000e+00 | 0.000000e+00 | 2022.0 | 36.619446 | -91.686566 |
| 25% | 1.267804e+07 | 533.000000 | 5.000000 | 9.000000 | 22.000000 | 1.153949e+06 | 1.859284e+06 | 2022.0 | 41.769168 | -87.710151 |
| 50% | 1.276897e+07 | 1033.000000 | 10.000000 | 24.000000 | 32.000000 | 1.167255e+06 | 1.893383e+06 | 2022.0 | 41.863073 | -87.661467 |
| 75% | 1.285712e+07 | 1731.000000 | 17.000000 | 35.000000 | 53.000000 | 1.176856e+06 | 1.910066e+06 | 2022.0 | 41.909023 | -87.626402 |
| max | 1.368384e+07 | 2535.000000 | 31.000000 | 50.000000 | 77.000000 | 1.205119e+06 | 1.951493e+06 | 2022.0 | 42.022548 | -87.524532 |
| Missing Values | Unique Values | Value Counts | |
|---|---|---|---|
| Case Number | 0 | 239573 | {'JF198311': 3, 'JF445443': 3, 'JF356096': 3, ... |
| Date | 0 | 112310 | {'01/01/2022 12:00:00 AM': 150, '08/01/2022 12... |
| Block | 0 | 27970 | {'001XX N STATE ST': 604, '0000X W TERMINAL ST... |
| IUCR | 0 | 306 | {'0810': 20111, '0820': 18885, '0486': 18692, ... |
| Primary Type | 0 | 31 | {'THEFT': 54888, 'BATTERY': 40946, 'CRIMINAL D... |
| Description | 0 | 286 | {'SIMPLE': 27226, 'OVER $500': 20111, '$500 AN... |
| Location Description | 972 | 135 | {'STREET': 67643, 'APARTMENT': 45770, 'RESIDEN... |
| FBI Code | 0 | 26 | {'06': 54888, '08B': 33964, '14': 27247, '07':... |
| Updated On | 0 | 1566 | {'01/03/2023 03:40:27 PM': 227123, '11/15/2023... |
| Location | 4735 | 118346 | {nan: 4735, '(41.976290414, -87.905227221)': 3... |
Note that only some of the variables from the whole datasets are used for our individual analyses. Since the data is prepared differently for answering each question, see details about cleaning and preparation in the next section.
4 Exploratory Data Analysis
4.1 Analysis 1: How does access to Academic Support Programs change when considering income levels and minority statuses?
By Giovanni Cacciato
To investigate accessibility to Academic Support Programs in the My Chi. My Future. directory based on income level and minority status, I used a Chicago Public Schools dataset from the 2020-21 school year. In cleaning this dataset, I dropped 2 virtual schools and 3 schools with student counts listed as zero. Additionally, I dropped two Academic Support Programs missing latitude-longitude pairs. In order to assess accessibility, I counted the number of in-person after school recurring Academic Support Programs offered for grade school-aged students (ages 5-21) within a mile radius of each school in the data (Near Programs), then analyzed this number for each school based on the percentages of low income and minority students.
It is important to note that the correlation between the percentage of low income students and the percentage of minority students is very positively strong (r≈0.86) due to historical disadvantages and lack of social support overall (see figure below), so I expected similar results among the two factors when looking at the accessibility of Academic Support Programs.
/var/folders/87/c47mf2zd1ts1w4gh05bwhwbc0000gn/T/ipykernel_11387/2294759022.py:53: UserWarning: Could not infer format, so each element will be parsed individually, falling back to `dateutil`. To ensure parsing is consistent and as-expected, please specify a format.
mychi['Start Time'] = pd.to_datetime(mychi['Start Time']).dt.hourText(0.5, 1.0, 'Percent Minority v. Percent Low Income in CPS')
Further, I hypothesized that there would be a negative trend in accessibility to Academic Support Programs for a rise in percentage of low income students, and a similar trend for a rise in percentage of minority students. So, I plotted the number of Near Programs against the Percent Low Income and Percent Minority in order to visualize my hypothesis. However, it was difficult to observe any trends that suggested this correlation on the surface level (see below figures). In fact, the plots show the opposite; a weak positive correlation suggesting that as schools have higher percentages of low income and minority students, there are more Near Programs.
Text(0.5, 1.0, 'Percent Low Income v.\n Number of Near Programs')
This was the opposite of what I expected, so I decided to dig a bit further and break the Percent Minority variable up by individual race. In doing so, I was able to see disparities in access to Academic Support Programs, specifically among schools with higher percentages of Hispanic students, and higher percentages of Native American students. Having weak negative correlation coefficients r≈-0.231 and r≈-0.046 respectively, a potential gap in accessibility to Academic Support Programs is suggested.
Although the correlation coefficients are not strong enough to make a definitive claim about the association between the number of Near Programs and Percent Hispanic/Percent Native American students, they did prompt me to look deeper into the accessibility of these programs, this time based on Percent Low Income and 4 different school types: Elementary Schools (ES), Middle Schools (MS), High Schools (HS) and Charter Schools. I found that among High Schools with a majority hispanic student population, the number of Near Programs slightly decreases as Percent Low Income increases (r≈-0.166).
Text(0.5, 1.0, 'Percent Low Income v. Number of Near Programs Among High Schools with a Majority of Hispanic Students')
Given the fact that among Chicago Public High Schools, roughly 45.1% have Hispanic students as their majority (see figure below), this is potentially a huge gap in Academic Support programming for this important student population.
Text(0, 0.5, '')
Despite there being signs suggesting some possible gaps in Academic Support Programs specifically for Hispanic high school students in the Chicago area, the data also suggest that many schools with lower percentages of low income students (i.e., higher percentage of students who do not fall into the low income category) tend to have less access to Academic Support Programs. Additionally, schools with higher percentages of white students tend to have less access to Academic Support Programs, as well (see Appendix). Although these aren’t exactly issues that we are seeking to address for stakeholders in terms of equitability, it is interesting to see.
4.2 Analysis 2: How are bilingual/multilingual communities represented in the opportunities (events/resources/programs) taking place in Chicago?
By Alexa Nuñez Magaña
To examine if the distrubution of bilingual/multilingual opportunities was representative of the Chicago population, we first looked at the percetage and behavior of bilingual/multilingual opportunities compared to the English monolingual opportunities. In order to do this, the My Chi. My Future dataset was subsetted to only include bilingual/multilingual opportunities using text analysis with a list of keywords/keyphrases that would signal that a program is bilingual/multilingual and checking if the title and/or description of an opportunity contained one of said keywords/keyphrases. Additionally, duplicate opportunities were droped as the programs’ categories were not relevant to this analysis.
We looked at the percentage of bilingual/multilingual vs monolingual opportunities in the dataset and compared it to the percentage of the Chicago population that speaks a language other than English at home given by the 2023 US census data.
This visualization allows us to see the striking difference between the percentage of bilingual/multilingual opportunities and the actual bilingual/multilingual population in Chicago. To look further into this gap we wanted to look at the amount of new bilingual/multilingual opportunities per year.
([<matplotlib.axis.XTick at 0x2c57f01a0>,
<matplotlib.axis.XTick at 0x2c57f1ca0>,
<matplotlib.axis.XTick at 0x2984e8410>,
<matplotlib.axis.XTick at 0x2c2db1280>],
[Text(2021, 0, '2021'),
Text(2022, 0, '2022'),
Text(2023, 0, '2023'),
Text(2024, 0, '2024')])
As we can see, the amount of bilingual/multilingual opportunities has steadily increased over the years. Although this is a promising trend to create a more representative number of opportunities for the Chicago population, we wanted to take a closer look at 2024, the year with the most number of new bilingual/multilingual opportunities, and analyze whether or not there were any changes gap found above.
Although the gap between bilingual/multilingual opportunities and the bilingual/multilingual population of Chicago is smaller when only looking at 2024, said gap is still substantially large. This starting analysis suggests that there already is a lack of equitability in opportunities for the bilingual/multilingual population based on quantity alone. Yet, We wanted to further explore the distribution and equitability of the existing bilingual/multilingual opportunities.
In order to achieve this analysis, we looked into the language demographics of each Chicago neighborhood using an interactive map with the help of the “Language Spoken at Home and Ability to Speak English” table from the 2015-2019 American Community Survey. To get a better analysis of the data an ‘other/unspecified’ column was added that included the difference between the sum of all the columns of languages spoken at home and the total population of the neighborhood. Additionally, a column including the amount of the population that did not speak only English at home was added to the dataset which was then used to compute the poportion of the non-English-only population per neighborhood. Lastly, a column including the most spoken language other than English was added for each neighborhood.
As for the My Chi. My Future dataset, since this analysis was dependent on the geographical distribution of the opportunities, online opportunities were omitted from this part of the analysis. Because the data was not categorized into compatible neighborhood labels as the “Language Spoken at Home” dataset, we used the latitude and longitude columns to find the location of each opportunity based on the coordinate reference system of the “Language Spoken at Home” dataset, since there were no missing values for either the latitude or longitude for any of the observations no further problems were encountered. Additionally, we wanted to look at the most offered language for the bilingual/multilingual opportunities of each neighborhood for which we created a “language” column that stored the languages other than English mentioned in the title or description of each opportunity. We then grouped by the neighborhood and opportunity languages to get the language with the maximum amount of counts. Lastly by grouping by “geographic cluster” (neighborhood) and “start year” we got the average amount of opportunities per year on each neighborhood.
/var/folders/87/c47mf2zd1ts1w4gh05bwhwbc0000gn/T/ipykernel_11387/3565055223.py:43: SettingWithCopyWarning:
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead
See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
bilingual_person_ops['Geographic Cluster Name'] = gdf['community']
/var/folders/87/c47mf2zd1ts1w4gh05bwhwbc0000gn/T/ipykernel_11387/3565055223.py:55: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.
For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.
chicago_info['opportunities'].fillna(0, inplace=True)
/var/folders/87/c47mf2zd1ts1w4gh05bwhwbc0000gn/T/ipykernel_11387/3565055223.py:67: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.
For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.
chicago_info['opportunity language'].fillna('No opportunities', inplace=True)Make this Notebook Trusted to load map: File -> Trust Notebook
This map can help us uncover two problems with the equitability of bilingual/multilingual opportunities for the Chicago population. The first problem is that there are neighborhoods (such as East Side, West Elsdon, Hegewisch, etc.) with a very high percentage of bilingual population yet no offered bilingual/multilingual programs. The second problem can be found in neighborhoods such as Norwood Park and Firest Glen where the most language spoken other than English doesn’t match the most offered language for the opportunities. Thus, aside from the lack of bilingual/multilingual opportunities, there is also a misditribution of the existing programs which fail to both reach the neighborhoods with higher bilingual/multilingual populations while also failing to accurately represent the bilingual/multilingual linguistic need of some neighborhoods.
This last problem was of special interest to us as it raised the question of whether or not the linguistic distribution of the bilingual/multilingual opportunities matched the linguistic distribution of the bilingual/multilingual Chicago population. No additional columns were created to complete this analysis.
Comparing the two linguistic distributions we can confirm our previous suspicion that there was a lack of representation between the languages offered in bilingual/multilingual opportunities and the languages spoken in the Chicago population. To be more specific it seems that there is a lack of representation of the non-Spanish bilingual/multilingual population based on the gap between the percentage of non-Spanish speakers and the percentage of bilingual/multilingual opportunities in said language. Furthermore, a second problem is found, one that could cause a lack of accuracy in the previous analysis of the language distribution: 47.3% of the bilingual/multilingual opportunities failed to specify which languages were offered/used/invited. Appart from hindering the accuracy in the analysis of the language distribution of opportunities, this lack of specificity could cause hesitation towards the bilingual/multilingual population as they wonder if their language-pair will be represented in a space were there seems to be an overgeneralization of “bilingual/multilingual” as refering to “Spanish-English speaker”.
4.3 Analysis 3: Are areas with higher crime rates associated with fewer programs or programs of specific categories? What about youth-related programs?
By Danny Ramirez
First, we want to look at the relationship between program count and crime count within the grid cells. Next, we’ll look at the statistical significance of our regression coefficient.
Correlation: 0.17139145663437833
Coefficient: 0.07790756999264808
Standard Error: 0.016484965299981814
T-statistic: 4.725977190424175
P-value: 2.7426170534639027e-06We observe that there is a weak but positive correlation between the two. Our coefficient of 0.0779 means that for each additional program in a grid cell, the crime counts increase by approximately 0.0779 crimes on average. This positive relationship means that areas with more programs tend to have slightly higher crime rates. We then tested the coefficient with a t-test. The code accounts for the degrees of freedom, so although heteroscedasticity is assumed, the code accounts for if it is not. We calculated a t-statistic of 4.73, which led to a p-value of < 0.0001. We reject the null at the 5% level, meaning that the coefficient is statistically significant. To explain this relationship, high-crime areas might attract more programs (such as intervention programs, community-oriented programs, etc.). It’s also possible that factors such as population density, socioeconomic status, or pre-existing infrastructure of the grid cell might influence both crime rates and program availability.
Here, we just wanted to get a closer look at the scatterplot. To do this, we regenerate the scatterplot filtering out grid cells where the number of crimes exceeds 3000 and the number of programs exceeds 1500.
Next, we wanted to look at the relationship between the youth program count and crime count within the grid cells. Again, we’ll look at the statistical significance of our regression coefficient.
Correlation: -0.044430847998046225
Coefficient: -0.039523322116876
Standard Error: 0.04638811375581901
T-statistic: -0.852013994897952
P-value: 0.3947619573520682We observe that there is a very weak and negative correlation between the two. Our coefficient of -0.0395 means that for each additional youth program in a grid cell, the crime count decreases by approximately 0.0395 crimes on average. This negative relationship means that areas with more youth programs tend to have slightly lower crime rates. We then tested the coefficient with a t-test. The code accounts for the degrees of freedom, so although heteroscedasticity is assumed, the code accounts for if it is not. We calculated a t-statistic of -0.85, which led to a p-value of 0.3946. We fail to reject the null at the 5% level, meaning that the coefficient is not statistically significant. Based on this, there is no strong statistical evidence that areas with more youth-related programs have more or less crime rates. It is likely that there exists a confounding variable that explains the negative correlation between the two.
Lastly, we wanted to look at which program categories were most prevalent in high-crime areas compared to low-crime areas.
Here, we can observe that the most popular program categories for both high-crime and low-crime areas are sports + wellness, music & art, and reading & writing. There does not seem to be a pattern between what program categories are typically offered in low-crime areas compared to high-crime areas, but it should be noted that programs that are more likely to be outdoors (sports + wellness and nature) are more popular in low crime areas than high crime areas and academia programs (academic support, computers, science) are more popular in high crime areas than low crime areas.
5 Conclusions
We notice that there is a lack of bilingual/multilingual opportunities listed in the program directory, and that there is a lack of opportunities in general to the Hispanic population in the Chicagoland area. As the Hispanic population is also largely bilingual/multilingual, this shows an overall gap in accessibility to programs for the Hispanic population due to lack of bilingual/multilingual opportunities and/or lack of any opportunities altogether.
6 Recommendations to stakeholder(s)
We found that Academic Support Programs are not as accessible to high schools with higher percentages of Hispanic students. In order to correct for this gap in accessibility, we suggest collaborating with Chicago Public High Schools with majority Hispanic student populations in order to provide Academic Support Programs close in proximity or even taking place inside of the schools directly after school hours as many programs begin during the school day, heavily affecting their accessibility.
Limitations: One possible limitation of this analysis is that it does not take into account proximity to Academic Support programs based on where individual students live. It is in fact very likely that students go to school in one area of Chicago and live in a different area that would potentially have more or less access to Academic Support Programs. Additionally, this analysis does not take into account access to modes of transportation that would potentially close the physical distance gap between students and Academic Support Programs. Having Academic Support Programs close to schools, although convenient, is not the only way to ensure access to these programs for students who need them most, which, for the purposes of this study, are students of low income and minority status.
We also found that the bilingual/multilingual population in Chicago doesn’t have equitable access to opportunities that represent them based on a lack of bilingual/multilingual opportunities, specially non-Spanish opportunities, in addition to a misplacement of existing opportunities that fail to reach neighborhoods with high levels of non-English-only speakers and/or fail to represent their language prominence. One way to solve the lack of bilingual/multilingual opportunities is by taking advantage of cultural opportunities that already seek a potential bilingual/multilingual population (ex. Día de Muertos, Lunar New Year, etc.). Even if the organizers aren’t speakers of any languages other than English, by advertising their event as bilingual they are inviting the actual bilingual/multilingual population to provide said environment through their presence. Additionally, as many of these opportunities are offered by the Chicago Public Library there are many bilingual/multilingual resources offered by request. By predicting the linguistic identity of the target audience and making these resources available regardless of whether someone requested them or not, this can increase both the number of bilingual/multilingual opportunities and the participation of the bilingual/multilingual population as they would not need to request being accommodated. The third recommendation is to fix the found problem of the large number of bilingual/multilingual opportunities that fail to specify the languages that will be spoken. Our recommendation would be to include a ‘language’ category in the app where organizers can specify the languages used in their event, this would prevent lack of clarity for the target population and might help find a more accurate linguistic distribution of the existing programs.
Limitations: Let’s keep in mind that for the purposes of our analysis “bilingual/multilingual” was defined as someone that spoke a language other than or in addition to English at home. However, the datasets used did not make a distinction between the two, therefore there could be an inclusion of monolingual non-English speakers that were considered bilingual based on this analysis. Yet, the existence of bilingual/multilingual opportunities in the languages of these monolingual non-English speakers included in the analysis still serves them as it invites them to a space where their language is used to some extent.
Finally, we found that there was a negative regression coefficient between youth-program count and crime count in a respective grid cell. Despite the coefficient not being statistically significant, the negative coefficient suggests a potential protective effect. We recommend that there be further investment in youth-oriented programs in high-crime areas, as they can be beneficial so long as there is monitoring to evaluate their impact. Stakeholders should also focus on the quality, type, and location of programs instead of increasing the number of programs without strategic planning. As noted on the bar graphs, the most popular programs in high crime areas are sports + wellness, music & art, reading & writing, and science, all of which relate to some form of academia. Simply increasing the number of programs does not necessarily reduce crime as seen in the positive, statistically significant coefficient, but more programs that are strategically placed and encourage academia or intervention-type programs can be beneficial in reducing crime.
Limitations: A possible limitation of this analysis is that the code does not primarily look into the impact of youth-related crimes when comparing them to youth-programs. It is possible that there would be a stronger correlation between the two and a more impactful regression coefficient because of the interventive nature of youth-programs because the coefficient was not statistically significant. The analysis also does not consider the socioeconomic status, population density, or pre-existing infrastructure of the location, which could influence both crime rates and program availability.
Appendix
Giovanni Cacciato
Text(0.5, 1.0, 'Percent of CPS White Students v.\n Number of Near Programs')